GitOps using Argo CD - Part1

In this article we will see how to use Git and Argo CD to deploy/ manage applications on your Kubernetes cluster. Before that, what is GitOps? Simply put, Operations driven using Git and CD tools! Following are the major components of GitOps:

• Infrastructure as code (IaC)
• Merge/ pull requests as change agent
• Continuous Delivery tool (Example: Argo CD)

Basically you keep all your application deployment manifests in a Git repository. And, if you like to make changes to your application, you create a merge/ pull request. Once it is approved and merged to the main branch a continuous delivery/ deployment tool like Argo CD will identify that and deploys the latest change to the target Kubernetes cluster. Here I am using a Tanzu Kubernetes Cluster with 1 control plane node and 3 worker nodes. I will be deploying Argo CD as well as my application on this K8s cluster. 

❯ kubectl get node
NAME                                STATUS   ROLES                  AGE   VERSION
gc-control-plane-rhpmq              Ready    control-plane,master   34d   v1.21.6+vmware.1
gc-workers-kfx7q-589888f77b-692n5   Ready    <none>                 34d   v1.21.6+vmware.1
gc-workers-kfx7q-589888f77b-jfzrs   Ready    <none>                 34d   v1.21.6+vmware.1
gc-workers-kfx7q-589888f77b-xvjsh   Ready    <none>                 34d   v1.21.6+vmware.1

Lets create a namespace first. 

❯ kubectl create namespace argocd
namespace/argocd created

❯ kubectl apply -f https://gist.githubusercontent.com/vineethac/dafa5b47afd674a1a9f7be2ce773a2bd/raw/4591e837098043b8095a5c48614e1c94b5ca2b44/tkg-psp.yml
clusterrole.rbac.authorization.k8s.io/psp:privileged created
clusterrolebinding.rbac.authorization.k8s.io/all:psp:privileged created

Following yaml file will install Argo CD: 

❯ kubectl apply -n argocd -f https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml

❯ kubectl get all -n argocd
NAME                                                    READY   STATUS    RESTARTS   AGE
pod/argocd-application-controller-0                     1/1     Running   0          20h
pod/argocd-applicationset-controller-5f7d8fffb7-82xgp   1/1     Running   0          20h
pod/argocd-dex-server-75f7cff9cd-7vc64                  1/1     Running   0          20h
pod/argocd-notifications-controller-69bf646f87-8bt5n    1/1     Running   0          20h
pod/argocd-redis-748569f956-bskfw                       1/1     Running   0          19h
pod/argocd-repo-server-8699756b5d-7qmx2                 1/1     Running   0          20h
pod/argocd-server-6dd9cd7964-gbfm4                      1/1     Running   0          20h

NAME                                              TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)                      AGE
service/argocd-applicationset-controller          ClusterIP   10.106.103.231   <none>        7000/TCP,8080/TCP            20h
service/argocd-dex-server                         ClusterIP   10.103.79.207    <none>        5556/TCP,5557/TCP,5558/TCP   20h
service/argocd-metrics                            ClusterIP   10.105.254.212   <none>        8082/TCP                     20h
service/argocd-notifications-controller-metrics   ClusterIP   10.97.254.140    <none>        9001/TCP                     20h
service/argocd-redis                              ClusterIP   10.97.244.161    <none>        6379/TCP                     20h
service/argocd-repo-server                        ClusterIP   10.101.181.242   <none>        8081/TCP,8084/TCP            20h
service/argocd-server                             ClusterIP   10.105.76.149    <none>        80/TCP,443/TCP               20h
service/argocd-server-metrics                     ClusterIP   10.100.168.241   <none>        8083/TCP                     20h

NAME                                               READY   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/argocd-applicationset-controller   1/1     1            1           20h
deployment.apps/argocd-dex-server                  1/1     1            1           20h
deployment.apps/argocd-notifications-controller    1/1     1            1           20h
deployment.apps/argocd-redis                       1/1     1            1           20h
deployment.apps/argocd-repo-server                 1/1     1            1           20h
deployment.apps/argocd-server                      1/1     1            1           20h

NAME                                                          DESIRED   CURRENT   READY   AGE
replicaset.apps/argocd-applicationset-controller-5f7d8fffb7   1         1         1       20h
replicaset.apps/argocd-dex-server-75f7cff9cd                  1         1         1       20h
replicaset.apps/argocd-notifications-controller-69bf646f87    1         1         1       20h
replicaset.apps/argocd-redis-748569f956                       1         1         1       19h
replicaset.apps/argocd-repo-server-8699756b5d                 1         1         1       20h
replicaset.apps/argocd-server-6dd9cd7964                      1         1         1       20h

NAME                                             READY   AGE
statefulset.apps/argocd-application-controller   1/1     20h

Lets port forward, so that you can access the Argo CD web UI. 

❯ kubectl port-forward -n argocd service/argocd-server 8080:443
Forwarding from 127.0.0.1:8080 -> 8080
Forwarding from [::1]:8080 -> 8080

Now you can access Argo CD in your web browser at https://localhost:8080/

username: admin
password: you can decode it from the following secret

❯ kubectl get secret argocd-initial-admin-secret -n argocd -o yaml
apiVersion: v1
data:
  password: TTFSRXJJZ0NKbHc2Y2JINA==
kind: Secret
metadata:
  creationTimestamp: "2022-06-22T13:00:14Z"
  name: argocd-initial-admin-secret
  namespace: argocd
  resourceVersion: "39224285"
  uid: d3b7c82e-0c92-4418-95eb-95a73fe674b6
type: Opaque

❯ echo TTFSRXJJZ0NKbHc2Y2JINA== | base64 --decode

Next step is to create a repository and add your application yaml files to it. You will also need to create a Argo CD application yaml file, push it to the repository, and then apply the same application yaml file to your cluster.

Following is a screenshot of my repo:

We have the application.yaml file and then inside the dev folder I have two yaml files.

 

 

 

Now, we also have the application yaml file. Make sure to paste your git repo url in the repoURL field.

 

Lets apply the application yaml manifest on to the Kubernetes cluster and that will connect Argo CD with your git repo.

Note: If you using a private repo, then you need to add your repository to Argo CD first and connect with respective credentials.

 

> kubectl apply -f /Users/vineetha/myrepo/argocd/application.yaml 

Once the application yaml is applied, after few seconds you can see nginx pod and svc getting deployed automatically under myapp namespace. 

❯ kubectl get pods,deployment,svc -n myapp
NAME                            READY   STATUS    RESTARTS   AGE
pod/my-nginx-74d7c6cb98-tzdfp   1/1     Running   0          2d

NAME                       READY   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/my-nginx   1/1     1            1           2d

NAME               TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)   AGE
service/my-nginx   ClusterIP   10.109.67.35   <none>        80/TCP    2d

 

 

From now on if you like to make changes to your application, say, you want to scale the replicas from 1 to 3, you have to create a merge request or pull request to the repository with the respective change, and once its approved and merged to the main branch Argo CD will automatically detect it and pull it and apply it to your Kubernetes cluster.

Hope it was useful. Cheers!

Reference video by Nana https://twitter.com/Njuchi_

Working with Kubernetes using Python - Part 04 - Get namespaces

Following code snipet uses Python client for the kubernetes API to get namespace details from a given context:

from kubernetes import client, config
import argparse


def load_kubeconfig(context_name):
    config.load_kube_config(context=f"{context_name}")
    v1 = client.CoreV1Api()
    return v1


def get_all_namespace(v1):
    print("Listing namespaces with their creation timestamp, and status:")
    ret = v1.list_namespace()
    for i in ret.items:
        print(i.metadata.name, i.metadata.creation_timestamp, i.status.phase)


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("-c", "--context", required=True, help="K8s context")
    args = parser.parse_args()

    context = args.context
    v1 = load_kubeconfig(context)
    get_all_namespace(v1)


if __name__ == "__main__":
    main()